3 Ways to Improve Forecast Accuracy with Process Improvement

Forecast accuracy vs process improvement is a conversation that begins early in the supply chain optimization process. When we begin conversations with a company, we often hear statements like “We need to improve our supply forecasts” or “We need a forecast model to tell us what our customers demand will be.”  These forecasts are critical components to any business team managing their supply chain or sales teams to be sure, but we’d suggest a slight tweak to your thought process that will frame both the problem and the solution in a more optimal way.

First, the uncomfortable truth: forecasts are wrong the moment they are created 99.9% of the time.  Why? Because forecasting is a process. The error rate, time spent, and usage of the forecasts will improve if:

1 – You have a forecasting and forecast maintenance process that fits your business

2 – You reduce time spent on forecast management

3 – Your forecasts integrate automatically with all ancillary systems which need them

Improving the forecasting process starts with understanding that many companies (especially commodities companies) have massive amounts of volatility on both the supply and demand sides of their business.  Volatility can cause decreased accuracy of forecasting whether it is created by the most knowledgeable person in your company or the best AI forecasting engine.  The more granular you attempt to make your forecast, the higher the potential error rate of that forecast will be.

Given these realities, optimizing your forecast granularity is critical.  For example, if you need to have a daily demand plan it is often better to first accumulate the historical demand by week across a group of similar products to decrease the granularity, then drive that weekly forecast down into a daily plan.  By doing this you reduce the noise in your data and reduce error rates in AI forecasts while still achieving the accurate daily plan that is needed.  Building the forecasting engine to fit your business is critical and will achieve far better results than grabbing a generic off the shelf tool or asking a salesperson or demand planner to forecast the demand they think they will achieve based on gut instinct.

Reducing the labor required by your team to forecast may be the most important part of a forecasting solution.  Salespeople and supply managers often avoid forecasting at all costs, seeing it as an enormous time sink that holds them to unreasonable expectations of accuracy. showing them how their time will be saved and forecasts will be accurate with this method will keep them from becoming disillusioned with the process and make them more likely to adopt the new strategy. 

Improving engagement by reducing time spent forecasting starts with a relatively accurate system-generated forecast and an interface which is easy to use while having all the contextual information they need right on the screen. Information like historical averages, volatility, contracts, etc. all make this process much faster which helps improve engagement from your team.

Finally, integration throughout the rest of the business systems reduces costly errors as well as the time forecast management requires for your organization. If you are currently using a spreadsheet process, as so many companies are, think about all the integration effort, copying and pasting into all the other systems that utilize the forecast, and time spent making sure the spreadsheets don’t break.  It often feels like the process is one miscue away from breaking down. The errors pile up in a process like that.  Not to mention the time, effort, and issues around measurement of forecast accuracy which is critical to accuracy improvement.

Creating a forecast in a system designed to be fully integrated to production planning, calculating what you have left to sell, etc., combined with improving the engagement with your sales and demand planning teams is the key to better forecasts.  It is the forecasting process, not the forecast that matters.

Want to see how AI-based data science and a company dedicated to customer happiness can help you improve forecast accuracy? Drop me a line.

You cannot copy content of this page